The work proposed here leverages detailed claims and enrollment data from a private health insurance exchange to learn more about the process of choosing a health plan when the choice set is large, and how this choice affects subsequent health care demand. My Plan by MedicaSM is an early example of a single-insurer private health insurance exchange. My Plan by Medica offers employers a choice set of up to 20 benefit plans from which employees select coverage, subsidized by the employer's defined-dollar contribution. An important enhancement was added for the 2013 plan year: in addition to selecting plans that vary by point-of-service cost sharing, employees now also select their provider network, choosing from the original broad PPO network or one of four distinct provider networks (Medica ACOs) built around vertically-integrated hospital and physician clinic systems. We propose to develop a model of this complex health plan choice, and examine how the resulting cost sharing and provider network characteristics impact health care demand in the years after that choice. Because this innovative Medica product incorporates differences in provider networks as well as point-of-service cost sharing, we believe the knowledge we gain from this model about health plan choice and its consequences can inform our understanding of the public exchanges to be offered under Affordable Care Act of 2010. The Specific Aims of the proposed work have been designed to support the AHRQ Value portfolio: Specific Aim 1: To identify the member characteristics that predict 2012 enrollment patterns, where options vary only by benefit design. Specific Aim 2: To identify the differences in 2013 and 2014 enrollment patterns, when Medica ACO options are layered on top of benefit design differences. Specific Aim 3: To identify the impact of benefit design differences on future patterns of care delivery. Specific Aim 4: To identify the impact of Medica ACO enrollment on future patterns of care delivery. The proposed work is innovative in the following ways: (1) We take advantage of access to detailed plan design, employee cost sharing, enrollment and claims data from the insurer to evaluate the impact of a private health insurance exchange put in place by an early entrant into this new market. The wide variety of employer groups assures variation in plan design and employee cost sharing. (2) Exogenous variation in employee out-of-pocket premiums provides an instrument to control for the non-random selection of plan design. (3) Our unusually rich data about provider networks allows us to use individuals' history of accessing primary care and specialty care provider networks to identify members' willingness to switch to a new usual source of care in order to save premium dollars. (4) We are able to follow members longitudinally to identify changes in patterns of care delivery associated with plan design.